Apply for class 03 now!

ConversationFit: Personalized on-device hearing aid fitting

About the project

Team Members

Alice Meier, Andreas Rabuser

License

MIT

Funding Period

In funding since 01/06/2026

What is the project about? 

Over 15 million people in Germany live with hearing loss. Hearing aids are configured using a one-size-fits-all approach, without taking individual needs into account. The test signals used by audiologists do not replicate real-world environments like restaurants, phone calls, or group conversations. Manual fitting is inefficient: users struggle to describe their problems clearly, audiologists cannot recreate these situations, and multiple appointments over several weeks are required. Older people, in particular, often struggle to articulate their problems. This is where ConversationFit comes in, using local speech-to-text and small LLMs to automatically detect comprehension issues. These are then converted into personalized, situation-specific adjustment recommendations.

Which audience does the project address?

The target audience consists of hearing aid users who have to live with suboptimal settings and thus suffer from unnecessary hearing difficulties in their daily lives. Particularly affected are people with OTC hearing aids who lack the typically time-consuming and costly professional fitting, those who find it difficult to articulate their fitting issues, and individuals who continue to have comprehension problems in specific situations despite using a hearing aid. Additionally, the goal is to support audiologists in providing data-driven recommendations.

What is to be achieved?

We are developing a privacy-first mobile app. The app will be cross-platform using Flutter and will offer local speech processing. We place a particular emphasis on data security and data sovereignty for users. Core components, such as speech analysis, are designed to be reusable so that they can also be used in a modular way for other health apps.